Application of Adaptive Control on Time-Varying Structure System

For improving the resistance of the structure to against the earthquake, the development of structural control becomes an important issue in civil engineering. Structural control can be classified into three types: Passive control, Active control and Semi-active control. The definition of Active control and Semi-active control is using the device installed in the structure which could adjust the behavior by input order to change the dynamic characteristic of the structure. Therefore, applying an ideal control algorithm to compute the input order is the core of the structural control technique.
LQR control is one of the significant control algorithms. The design of the LQR controller is depend on the system’s information. However the control effeteness will be poor when the system is time-varying. The purpose of this research is trying to improve the adaptive ability of the LQR controller by combining the system identification technique. And the controller could update the system information of the time-varying system to reform the original LQR controller’s effeteness.
The research designed an adaptive controller which is different from the original LQR controller for time-varying system. First in Chapter 2 will introduce the LQR algorithm and to verify the control superiority by a 2 DOFs connected structure control experiment. Then in Chapter 3 we will mention about the system identification algorithm: Subspace Identification which could identify the system information including the natural frequencies and stiffness matrix precisely. In Chapter 4 we combined these two algorithms and design two methods to create the adaptive controller. One of the methods is System Compensator which is used to compensate the reduction of the system’s stiffness by increasing the control force and make the time-varying system behaves like a healthy system. Another method is Gain Switcher which will switch an optimal control gain corresponded to the level of stiffness reduction of the time-varying system. Finally, we used Simulink to simulate the time-varying systems responses under these two adaptive control methods with the mathematic model of MR damper to verify the effeteness. By updating the system’s information using subspace identification, the two adaptive controllers reduced the displacement responses ideally.